How accurate is genomic prediction across wild populations?

Area of Science:

• Evolutionary ecology
• Genomics
• Quantitative genetics

Background:

• Genomic prediction, predicting genetic values from marker data, has transformed breeding and medicine.
• Its application in wild populations, especially across different groups, is less explored.
• Across-population genomic prediction is crucial for understanding local adaptation and spatially varying evolutionary processes.

Purpose of the Study:

• To apply genomic prediction across wild house sparrow populations.
• To compare the accuracy of across-population versus within-population genomic prediction.
• To understand the mechanistic basis of across-population prediction accuracy by linking it to population differentiation.

Main Methods:

• Genomic prediction models were trained and tested across distinct wild house sparrow populations.
• Prediction accuracy was compared between models trained and tested within the same population and those trained in one population and tested in another.
• Relationships between across-population prediction accuracy and population differentiation metrics were analyzed.

Main Results:

• Predictions across populations were consistently less accurate and more variable than within-population predictions.
• Across-population prediction accuracy showed a covariation with specific population differentiation measures.
• The study identified limitations in current theories explaining genomic prediction accuracy across populations.

Conclusions:

• Across-population genomic prediction in wild populations is challenging, with accuracy dependent on population structure.
• Understanding the mechanisms driving prediction accuracy across populations is essential.
• Novel methods are needed to effectively utilize genomic data for studying spatially explicit evolutionary processes.